Signal processing and feature extraction toolbox for prognosis/diagnosis

In this paper we describe a novel general toolbox that we have developed as part of the integrated prognosis and diagnosis framework. This tool can be used to select the appropriate signal processing techniques for enhancing a signal, to select a portion of a signal, to represent a signal in 2 dimensional domains such as the time-frequency, to extract appropriate prognostic/diagnostic features either directly from the signal or from its representations and to generate a required sensor's signal that would be useful in prognosis/diagnosis. This toolbox is distinct from the Matlab's signal processing toolbox or any other commercially available signal processing toolbox in that (a) it provides techniques suitable for prognosis/diagnosis at the same time keeping the generality like the Matlab's Signal processing toolbox. In addition, (b) it provides the facility to save the optimized algorithms to use it during the run time of the prognosis of a system. Our toolbox mainly consists of three GUIs: (1) the signal processing/feature extraction (2) the window and (3) the coefficients. These GUIs are designed such that they can be operated either individually or made to interact with each other. In particular, the signal processing/feature extraction GUI can be independently used to synthesize any sensor signal that is of prognosis or diagnosis value but is not available since that sensor was not built in the system that is to be diagnosed. The details on the functionalities of each of these GUIs are provided in this paper.